“Deep Mapping” with Dr. Mike Sawada Tuesday, March 20th in Ottawa
Please come and join us on Tuesday, March 20th at the RA Centre, Outaouais Room, for a very interesting and informative presentation from Dr. Mike Sawada, Professor, Department of Geography, Environment and Geomatics, University of Ottawa. Mike’s presentation entitled “Deep Mapping” will discuss the use of Google street view imagery to obtain ground-level perspectives with sufficient temporal resolution for automating and studying urban structural changes. This can facilitate automated structural classification and qualitative structural changes in the built environment in Ottawa.
Advancements in deep learning make news daily with references to self-driving cars, check-out free retailing and apocalyptic visions of the ‘rise of the machines’. From a remote sensing perspective, there exists a vast underutilized source of imagery at the street level, collected by Google Inc. over the past decade in North
America. Street-level imagery provides a ground-level perspective with sufficient temporal resolution for studying urban structural changes. Convolutional neural networks can take these images as input and are capable of learning high-level feature representations. We have been using deep convolutional neural network models to automate the mapping of various features of the built environment that are difficult to discern from air/spaceborne nadir images. The presentation will focus on our use of deep learning to mine Google Street View imagery for purposes of automated structural classification and qualitative structural changes in the built environment in Ottawa. The construction and application of our models will be discussed.
For more information about the presentation and to register, please click on the link CIG Ottawa Branch Homepage